Marit Econ Logist DOI 10.1057/s41278-017-0077-5 ORIGINAL ARTICLE

On the cost of ice: estimating the premium of Ice Class container vessels Tomi Solakivi1 • Tuomas Kiiski1 • Lauri Ojala1

 Macmillan Publishers Ltd 2017

Abstract Cargo vessels navigating ice-infested waters have to comply with additional requirements including a strengthened hull and additional engine power, modifications that affect their operational economics. This paper analyses the additional cost of ice imposed on container vessels conforming to the most commonly used Finnish-Swedish Ice Class Rules. Cost estimations are based on descriptive statistics and regression analysis. The aim is to assess the reportedly higher costs of IA and IA Super Ice Class vessels compared with lower-classed or conventional vessels. Our analysis reveals that Ice Classed vessels incur significantly higher fuel and capital costs. This increases the shipping costs of Ice Classed vessels in open water by 9% (1 USD/TEU/day) compared with the other vessels under review. In wintertime, Ice Classed vessels unit costs could be up to 4 USD/ TEU/day higher than for vessels in the reference group in open water. Our findings contribute to the discussion on the economics of shipping in winter conditions, and on the imposition of environmental regulations in the industry when operating in ice-infested waters. These themes have been scantly covered in extant literature, but are important for countries with trade depending on winter navigation. Keywords Ice Class fleet  Winter navigation  Container shipping  Cost structure  Cost differentials

& Tomi Solakivi [email protected] 1

Operations and Supply Chain Management, Turku School of Economics, University of Turku, 20014 Turku, Finland

T. Solakivi et al.

Introduction Between 1948 and 2013, the global value of merchandise trade increased by a factor of 200 (WTO 2015). Major contributors to this included the relative decline in transport costs through improved efficiency in terms of equipment and information flow, as well as trade liberalisation. The location of production also had an effect. Transport costs as such alter relative prices and patterns of trade (Hummels 2007) by making new regions competitive at an international level (Krugman 1991). In terms of value, almost 80% of world trade is transported by sea, of which container vessels ship around 60% (World Shipping Council 2015). According to UNCTAD (2016), global seaborne trade totalled 10,047 million tonnes in 2015. Of this, about 15% by volume was containerised cargo, having had increased by a factor of 10 since 1985. The growth in container traffic was two-to-three times more rapid than in any other type of seaborne cargo. Many container lines operate in seasonally ice-infested waters and therefore require specialised Ice Class vessels. The areas in question include, but are not limited to the Baltic Sea and routes between Continental Europe and Greenland. Maersk, for example, has ordered seven Ice Class 3600 TEU (twenty-foot equivalent unit) container vessels to be deployed in the Baltic and the North Sea from 2017 (Lloyd’s Register 2015; Maersk 2015). The area in which seasonal ice is encountered is unevenly spread in the Northern Hemisphere. Most—if not all—of the ports of countries such as Finland and Estonia tend to be ice-bound during the winter (Omstedt et al. 2014). Some parts of Canada (e.g. St. Lawrence Bay and the Labrador Sea), Sweden (the Gulf of Bothnia) and Russia (the Bay of Finland, Barents Sea and the Sea of Okhotsk) are also affected. Predictions regarding the increased accessibility of the Arctic Sea (Stephenson et al. 2013) make the question of maritime transport in ice-infested waters relevant, also for new countries and regions. As of January 2015, there were 358 container vessels in the worldwide Ice Class fleet, with technical features that deserve more attention (CRSL 2015). The design of these vessels is a trade-off between open-water performance and ice capability. Among other things, their hulls need to be strengthened, making them more expensive to build than conventional vessels (Erikstad and Ehlers 2012). They also need additional engine power to cope with icy conditions, and this affects fuel consumption, as well as their newbuilding price (Lasserre 2014). Designed to operate in ice-infested waters, Ice Class vessels must comply with the relevant requirements in open-water conditions. Conditions in the Baltic Sea are ideal for about three months in the winter. However, given their sub-optimal design for open-water conditions, they incur extra costs during the rest of the year. These additional costs are likely to affect the relative prices of goods traded by countries and regions relying on shipping in ice-infested waters. This research analyses the additional ‘‘Cost of Ice’’ of container vessels, conforming to requirements as defined in the Finnish-Swedish Ice Class Rules (FSICR). Recently available calculations tend to neglect these additional costs, or to rely on heuristic industry estimates with limited generalisability in an Arctic context

On the cost of ice: estimating the premium of Ice Class…

(Xu et al. 2011; Pruyn 2016). Thus, existing cost estimates may not cover the real costs of shipping in icy conditions. Moreover, given that environmental effects and costs are based on similar methodologies (e.g. Corbett et al. 2009), the environmental effects of shipping in winter conditions may also be underestimated. The purpose of this research is to assess the excess capital, operating and fuel costs of IA and IAS Ice Class (FSCIR) vessels compared with conventional vessels. Although limited in number, Ice Class container vessels were chosen for analysis, given that containers are dominant in the transportation of manufactured goods. Moreover, the market for such goods is heavily exposed to international competition and changing trade patterns. This paper contributes to the extant literature in numerous ways. First, it complements existing knowledge in presenting the additional costs incurred by Ice Class container vessels, compared to conventional vessels operating in open-water conditions. Second, the analysis provides a baseline on which to calculate the true costs of operating in icy conditions. Third, cost estimates for operating in winter conditions provide a basis on which to evaluate the viability of Arctic Sea routes. Fourth, more precise estimates of the fuel consumption of Ice Class vessels in summer and winter conditions enhance understanding of the environmental impact of shipping. The paper continues as follows. Earlier literature on Ice Class and the costs of shipping is reviewed in the following section, after which the research methodology and data sources are described. The empirical results are presented next, and the final section sets out and discusses the conclusions.

Ice Class and the costs of shipping Ice Class notations are technical standards for vessels navigating ice-infested waters. In principle, the Ice Class is determined in accordance with the ice conditions and the necessary safety level related to the type and area of operation. In practice, however, Ice Class depends on the requirements of maritime authorities. FSICR is the industry standard used in designing ships for seasonal ice conditions (see e.g. Finnish Transport Safety Agency 2011; Riska and Ka¨ma¨ra¨inen 2011). Most classification systems adopt FSICR standards, and there is rough equivalency between the respective rules. Ice-capability requirements reflect conditions resembling those in the Northern Baltic Sea in winter, where icebreakers escort vessels in difficult conditions (Goerlandt et al. 2016). The FSICR comprises four Ice Classes (IA Super, IA, IB and IC), including notations for non-strengthened vessels (Classes II and III). The requirements relate to the structural strength of the vessel’s hull, machinery and overall performance through ice. The hull must withstand the pressure exerted by the ice. Ice performance is also linked to minimum engine power requirements for navigating in brash-ice channels at a minimum speed of five knots. For example, IA Super (IAS) vessels are capable of operating in channels covered by a 10-cm-thick layer of consolidated ice. Other regulatory regimes include the Russian Maritime Register of Shipping and the Polar Class rules of the International Association of Classification Societies. These usually apply in more severe ice conditions in Arctic waters.

T. Solakivi et al.

Regression analysis has been used in several studies to quantify the costs of shipping (Cullinane and Khanna 1999; Tran and Haasis 2015). Shipping costs are usually divided into three categories: (i) capital costs, depending on the purchase price of the vessel; (ii) operating costs, including crew wages, maintenance and insurance; and (iii) voyage costs, including fuel and other, voyage-related expenses, such as ice-breaking fees. However, there is limited knowledge about the costs of shipping in winter conditions. Before the recent emergence of interest in Arctic shipping, fuelled by climate change (Lasserre 2014), only a few studies had focused on Ice Class vessels (Koskinen and Hilmola 2005). It appears though that Ice Class requirements and the more challenging operational environment increase shipping costs. Previous estimates of the additional costs of building Ice Class vessels vary between six and 30% for the FSICR Ice Classes IC-IAS (Laulajainen 2009; Liu and Kronbak 2010; Erikstad and Ehlers 2012). The manning costs of vessels operating in the Arctic have been estimated to be 10% (Liu and Kronbak 2010; Lasserre 2014), and maintenance costs 20–100% (Liu and Kronbak 2010; Lasserre 2014) higher than the costs of conventional vessels. Similarly, insurance premiums vary from 25 to 100% for Protection and Indemnity, and between 30 and 100% for Hull and Machinery (Liu and Kronbak 2010; Pruyn 2016). The specialised design of an Ice Class vessel, in terms of bow shape and a heavier reinforced hull, affects fuel consumption. This effect is estimated to increase consumption in a range from less than one to 6% depending on the Ice Class (Dvorak 2009; Erikstad and Ehlers 2012). Whereas previous design effects on fuel consumption are well documented, the additional engine power required by ice class vessels (Riska and Ka¨ma¨ra¨inen 2011) has not been analysed. Ice-breaking fees also usually depend on Ice Class (e.g. Gritsenko and Kiiski 2016). Moreover, the increased hull weight reduces a vessel’s cargo-carrying capacity (Laulajainen 2009), although this is thought to be of less significance for container ships (Erikstad and Ehlers 2012). There is notable variation in the estimates of cost additions in the literature. One reason for this could be the use of industry estimates that are hard to compare as the sources and methodologies are inadequately reported (Pruyn 2016). What makes it even more complicated is that Ice Class vessels operate in icy conditions for only a small part of the year. The Baltic winter lasts for an average of three months (Omstedt et al. 2014), for example. For the rest of the year, vessels operate in open waters, but they still have to meet Ice Class requirements.

Methodology Following the example of Cullinane and Khanna (1999), this research analyses costs at sea in three categories: (i) capital, (ii) operating and (iii) fuel costs; together they constitute around 20% of the total transport costs in a typical transport chain (Tran and Haasis 2015). The costs of vessels conforming to the two highest FSICR classes (IA or IAS) are compared to those of vessels in the lower class. IA and IAS vessels

On the cost of ice: estimating the premium of Ice Class…

are meant for year-round navigation in the Baltic Sea (Riska et al. 1997), but are also suitable for Arctic Sea routes (Erikstad and Ehlers 2012). In calculating fuel costs, Cullinane and Khanna (1999) multiply the estimated daily consumption of fuel oil by the unit price per tonne. The following equation is used for daily fuel oil consumption (FO): FO ¼ Installed kW  SFOC  Engine load ð80%Þ 

24 ; 1;000;000

ð1Þ

where Installed kW is referring to the main engine power of the vessel, and SFOC to the Specific Fuel Oil Consumption of the main engine. In our paper, the vessel size elasticity of Installed kW is estimated by means of regression analysis. The starting model for installed power is lnðkW Þ ¼ a0 þ a1 lnðTEU Þ þ b1 Ice Class þ

2 X

c1 Timei þ e;

ð2Þ

i¼1

where Ice Class = 1 if the vessel is classified as IA or IAS, and 0 otherwise. Two years are represented with two dummy variables: Time1 = 1 if the vessel was built in 2000 and 0 otherwise, and Time 2 = 1 if built in 2005, and 0 otherwise. In these years, changes in the FSICR technical requirements have had a significant impact on the power requirements of Ice Class vessels (Riska and Ka¨ma¨ra¨inen 2011). Table 1 summarises the variables used in the analysis. The regression model was estimated by the General Linear Model with backward elimination, and the p value as a criterion. Variables were dropped one by one based on their p value, until only the significant ones were left. Data on the installed power of fully cellular container vessels (FCC), taken from Clarkson’s World Fleet Register (CRSL 2015), were used to estimate installed engine power in kW. The main-engine Specific Fuel Oil Consumption (SFOC) was also calculated from data obtained from the same source. A concern related to calculating fuel consumption is that Ice Class vessels utilise both ends of the main-engine load spectrum depending on the operational conditions. The load is very low in open waters, and (very) high in ice conditions (Man Diesel and Turbo 2013). However, in our calculations, it was assumed to be constant at 80%, as has previously been done (see Corbett et al. 2009). To minimise the effect of fuel-price volatility, calculations were based on the five-year average (2010–2015) of two grades of Intermediate Fuel Oil (IFO), IFO180 and IFO380 (Bunker Index 2015). The per-tonne price of IFO180 varied Table 1 Variables used in the analysis

Variable

Name

Scale used

Installed power

kW

Scale variable; 0–131,433 kW

Vessel size

TEU

Scale variable; 0–19,224 kW

Ice Class

Ice Class

1 = IA and IAS; 0 = Other

Time (year)

Time 2000

1 = before 2000; 0 = after 2000

Time 2005

1 = before 2005; 0 = after 2005

T. Solakivi et al.

between USD 312 and USD 752, with a five-year average of USD 599. The corresponding figures for IFO380 were USD 358 and USD 786, with a five-year average of USD 629. The model for vessel prices in this research follows the example of Cullinane and Khanna (1999). As with engine power, dummy variables were included, in addition to vessel size, to hopefully capture the effect of the years 2000 and 2005 (see above). Hence the following model: lnðPriceÞ ¼ a0 þ a1 lnðTEU Þ þ b1 IceClass þ

2 X

c1 Timei þ e:

ð3Þ

i¼1

Data from Clarkson’s World Fleet Register (CRSL 2015) were used to estimate vessel price, limited to the newbuilding price of fully cellular container vessels. Prices are expressed in US dollars, while prices in other currencies were converted to US dollars according to the exchange rate parity of the respective year (Federal Reserve 2016). The newbuilding prices are presented in Fig. 1 as a price-per-TEU ratio. As with the model estimating installed power, a General Linear Model with backward elimination (Draper and Smith 1998) was used for the price estimation. Cullinane and Khanna (1999) calculated the daily capital costs, expressed as an annuity for 20 years (the assumed economic life of the vessel), a 10% interest rate and a residual value of zero. Given that current interest rates are significantly lower, and as recommended by Wijnolst and Wergeland (2009), the Libor (London Interbank Offered Rate) rate was used, plus a margin of 1.5%. To avoid interest-rate volatility, calculations were based on a five-year average (2010–2015) of the 12-month Libor rate from the Federal Reserve Bank of St. Louis. Furthermore, a 25% residual value was assumed for the vessel (Wijnolst and Wergeland 2009). Operating costs were obtained from Drewry Maritime Research: Ship Operating Costs Annual Review and Forecast (Drewry 2012). Cost data covered manning, insurance, stores, spares, lubes and the R&M and M&A costs of container vessels in

Fig. 1 Newbuilding prices of Fully Cellular Containerships per TEU in million USD

On the cost of ice: estimating the premium of Ice Class…

seven size categories. These ranged from 500–700 TEU to 10,000–12,000 TEU, and within each range operating costs were assumed to be constant. Operating costs, including manning, maintenance and insurance, tend to increase with the level of risk associated with the operating environment (Lasserre 2014). Consequently, no empirical data were available to statistically analyse the exact cost differential. For this reason, operating costs were assumed to be the same for both IA and IAS class and other vessels. Hence, our model results are based on fuel and capital cost differentials.

Results Descriptive statistics Data on capital and fuel costs were obtained from Clarkson’s World Fleet Register (CRSL 2015). The register covers all vessels, operational or on orderbook, up to January 2015. The dataset contained technical data on 93,930 vessels, of which 5533 were FCCs. Of these, 5096 were currently operational, and 437 were on order. The majority of them (4170) were built after 2000; 270 between 1990 and 1999; and 198 before 1990. The dataset contained information on 358 container vessels equivalent to Ice Class IA or IAS, and 5175 vessels with an Ice Class below IA (Table 2). The newbuilding price information was available for 1701 FCCs, 40 of which had an Ice Class of IA or IAS. Data on main and auxiliary engine power were available for 5349 and 1401 vessels, respectively, of which 353 and 87, respectively, had an IA or IAS Ice Class. Main-engine SFOC was available for 3565 vessels and auxiliary engine SFOC for 435 vessels. As expected, vessels with a high Ice Class comprise a small part of the total fleet. In addition, those classed IA or IAS are smaller than the other vessels. Of the 358 high Ice Class vessels, 338 had a TEU capacity of less than 2000 (Table 3), against 2000–4000 in the remaining 20. The largest currently operational container vessel (in 2015) in Ice Class IA has a capacity of 3534 TEU. For this reason, the dataset was first analysed as a full set. Second, all Ice Class and Non-ice Class vessels exceeding 4000 TEU were excluded from the regression, to see if the cost functions behaved differently. Table 2 Availability of technical data in the database, as of January 2015 (CRSL 2015)

Ice Class IA or IAS Fully cellular container vessels Newbuilding price Main-engine power (kW)

Ice Class IB, IC, ID or no Ice Class

358

5175

40

1661

353

4996

Auxiliary engine power (kW)

87

1314

Main-engine SFOC (g/kWh)

156

3409

5

430

Auxiliary engine SFOC (g/kWh)

T. Solakivi et al. Table 3 Division of container vessels into size categories according to Ice Class, as of January 2015 (CRSL 2015)

TEU

Ice Class IA or IA S

Other Ice Class

Total

1–1999

338

2019

2357

2000–3999

20

980

1000

4000–5999

–

1007

1007

6000–7999

–

284

284

8000–9999

–

530

530

10,000–11,999

–

54

54

12,000–13,999

–

175

175

14000–

–

126

126

Fuel costs Two variables were estimated with respect to fuel costs. First, the SFOCs of the main and auxiliary engines were measured in g/kWh, estimated from Clarkson’s World Fleet Register (CRSL 2015). The average main-engine SFOC was 172.5 g/ kWh, compared with 252 g/kWh for the auxiliary engine. ANOVAs were run to assess possible differences in SFOC between IA and IAS vessels and the control group, which turned out to be insignificant. Consequently, the estimated averages were used for calculating the fuel consumption and fuel costs of the vessels. Second, two regression lines were fitted to calculate the fuel consumption of the main engine; one with vessels with less than 4000-TEU capacity, and the other with all the container vessels included. Table 4 presents the results of the regression analysis. The findings indicate a change in direction, since 2005, in the development of main-engine power in IA and IAS Ice Class as opposed to conventional vessels. Apparently, changes in the FSICR technical requirements have led to changes in engine power, and thus in fuel consumption and costs, for Ice Class vessels. The elasticity of the main-engine power with respect to the TEU capacity was estimated at 0.954 and 0.875, respectively, in the regressions using the limited and the full dataset. These estimates are in line with those of Cullinane and Khanna (1999), indicating an elasticity of 0.967 for container vessels with a capacity of less than 4000 TEU. However, our results differ from those of Tran and Haasis (2015), who reported an elasticity of 0.51 in container vessels larger than 4000 TEU. Our difference with previous results calls for deeper analysis. Consequently, a regression was also carried out for a sub-sample of container vessels with a capacity of more than 4000 TEU. The resulting elasticity coefficient was 0.49, with an R2 of 0.447. This indicates a wide variation in propulsion power and fuel consumption among vessels with the same TEU capacity. The propulsion power of vessels with a capacity of around 10,000 TEU (9500–10,500 TEU), for example, varies between 40,670 and 84,582 kW. Moreover, as with the newbuilding prices, the main-engine power (and thus fuel consumption) was lower (-0.093 and -0.063 in the limited and the full model, respectively) in vessels with an Ice Class other than IA or IAS. Figure 2 shows the daily fuel costs of container vessels up to 4000 TEU. The costs were calculated following the Corbett and Koehler (2003) example, under the

On the cost of ice: estimating the premium of Ice Class… Table 4 Regression coefficients of the main-engine power analysis TEU \ 4000

All vessels included

B

SE

t

B

SE

t

Intercept

2.567**

.041

62.744

3.120**

0.033

93.959

lnTEU

.954**

.006

166.377

0.875**

0.004

199.296

Ice Class = Other

-.094**

.013

-7.161

-0.040**

0.015

-2.603

Ice Class = IA and IAS

0a

–

–

Dependent variable: lnkW

Time 2005 = After 2005 Time 2005 = Before 2005 R2: 0.896

0a

–

–

0.019**

0.007

-12.96

0a

–

–

R2: 0.933

Breusch–Pagan test of Heteroscedasticity: sig. 0.145 Durbin–Watson test: 1.257 0a = Reference category ** Significant at 0.01 level

Fig. 2 The fuel costs of IA/IAS Ice Class and conventional container vessels up to 4000 TEU in USD per day, at a fuel price of USD 325 per tonne

assumption that the auxiliary engine power was 10% of the propulsion power. The load factor of the main-engine was assumed to be 80%. Fuel price (USD 325 per tonne) was calculated according to the 2010–2015 5-year average for IFO380 fuel in Rotterdam. Costs were around USD 42,100 for a vessel of 4000 TEU and an Ice

T. Solakivi et al.

Class of IA or IAS, and USD 38,300 for a similar-sized vessel of another Ice Class. Hence, there was a 10% difference in daily fuel costs. Capital costs Table 5 presents the results of the regression analysis with both the limited and the full dataset concerning the purchase price of container vessels. The elasticity of the newbuilding price with respect to capacity was 0.699, when the dataset only included vessels under 4000 TEU, and 0.739 when larger vessels were added. These estimates are well in line with previous results: 0.759, according to Cullinane and Khanna (1999), and 0.7 according to Tran and Haasis (2015). In addition, both models indicate a statistically significant difference in the purchase price of vessels depending on Ice Class. The new building prices of vessels with an Ice Class other than IA and IAS were -0.12 (-0.131), or around eight% lower than those of IA or IAS vessels. Figure 3 shows the daily capital costs of container vessels under the assumption of a 20-year life and a 25% residual value. The interest rate is assumed to be equal to the 5-year average of the 12-month Libor rate (0.78), with a margin of 1.5 percentage points. Operating costs Operating costs were obtained from Drewry (2012), and were assumed to be similar for all vessels regardless of the Ice Class. As with capital and fuel costs, operating costs were also modelled with a log–log regression. The elasticity of operating costs w.r.t. size (TEU) was estimated at 0.31 (Table 6).

Table 5 Regression coefficients of the newbuilding price analysis TEU \ 4000 B

All vessels included SE

t

B

SE

t

-29.708

Dependent variable: lnPrice Intercept

-1.682**

0.144

-11.708

-1.972**

0.066

lnTEU

0.699**

0.19

36.345

0.739**

0.008

98.374

Ice Class = Other

-0.120**

0.043

-2.754

-0.131**

0.040

-3.242

0a

–

–

0a

–

–

Ice Class = IA and IAS

R2: 0.669 Breusch–Pagan test of Heteroscedasticity: sig. 0.856 Durbin–Watson test: 0.941 0a = Reference category ** Significant at 0.01 level * Significant at 0.05 level

R2: 0.869

On the cost of ice: estimating the premium of Ice Class…

Fig. 3 Capital costs in USD/day of IA/IAS Ice Class and conventional container vessels Table 6 Regression coefficients of the operating cost analysis

TEU \ 4000 B

SE

t

Dependent variable: ln operating costs

** Significant at 0.01 level

Intercept

6.290**

.112

56.010

lnTEU

.311**

.013

23.477

R2: 0.905

Figure 4 shows the estimates of operating costs per day. By way of comparison, the operating costs of a vessel with a capacity of 4000 TEU were estimated to be around USD 7111 per day. Following the logic of previous authors (Tran and Haasis 2015), shipping costs were calculated as the sum of capital, fuel and operating costs. Figure 5 shows the shipping costs of container vessels of up to 4000 TEU. The estimated daily costs for a 2,000 TEU container vessel are between USD 31,500 and USD 34,200, depending on the Ice Class. The respective costs for a 4000 TEU vessel are estimated at USD 55,600 and USD 60,600. Overall, the difference in shipping costs between IA/IAS Ice Class and other vessels is around 8 or 9%. The higher purchase price of the vessel leads to higher

T. Solakivi et al.

Fig. 4 Total operating costs per day in USD per vessel

Fig. 5 Shipping costs (USD; left y-axis) and unit costs (USD/TEU; right y-axis) per day for IA/IAS Ice Class and conventional container vessels up to 4000 TEU

capital costs, and the higher propulsion power incurs higher fuel costs. The difference in costs is estimated to be USD 1.35 and 1.26 per TEU per day, respectively, for a 2000 and a 4000 TEU vessel. The costs per TEU per day for larger (e.g. 10,000 TEU) vessels (Xu et al. 2011) would be USD 1.16 higher for an Ice Class than for a conventional vessel.

On the cost of ice: estimating the premium of Ice Class…

The 9% difference is in itself a significant addition to the costs of transport. However, the estimate is based mainly on technical data, without taking into account winter conditions. Just as the sea margin adjusts calculations to account for the effects of rough weather and hull fouling, a ‘‘winter margin’’ could adjust for weather conditions during winter months. To assess the effects of winter conditions, therefore, 43 shipping companies, operating in the Baltic Sea, were interviewed in August 2015. Of these, 15 were able or willing to give an estimate of how much fuel consumption increases in winter conditions. Their estimates varied between 7 and 70%, depending on the vessel type and size, the average being 31.6% and the median 26.5%. The large variation in these estimates reflects the complexity of the conditions that prevail when navigating in ice (Bourbonnais and Lasserre 2015) and the severity of these conditions. The highest were based on activity requiring the ship to use more power to break brash ice and the lowest were based on stormy conditions. Even with this relatively small sample and indicative results, it would seem safe to assert that the fuel consumption of an average vessel increases by 25–30% in winter conditions. Combined with the 10% higher engine power, the estimated increase of 30% in fuel consumption would mean that the ‘‘Cost of Ice’’ could raise fuel costs by 40%. This means that the shipping costs of a 4000 TEU Ice Class IA/IAS vessel could be 30% higher than a conventional vessel of similar size operating in open water. Additional costs are incurred through reduced operating speed in wintertime. The interviewees estimated that winter conditions could reduce operating speed by an average 30%. Although a slower speed may temporarily lower fuel consumption, the overall cost burden facing Ice Class vessels for a given transport leg is substantially increased.

Conclusions and discussion Our analysis revealed a significant difference in newbuilding prices between the Ice Class vessels (IA/IAS) and the reference group. On average, the price of an Ice Class IA or IAS container vessel was 8% higher than that of a similar-sized vessel with a lower Ice Class. This is consistent with the findings of Erikstad and Ehlers (2012), but contradicts the previous higher estimates of Laulajainen (2009) and Liu and Kronbak (2010). The implication is that the capital costs of an Ice Class IA/IAS vessel are, on average, 8% higher than those of a conventional vessel. The elasticity of the newbuilding price of container vessels was around 0.7, which is in line with the previous findings of Cullinane and Khanna (1999) and Tran and Haasis (2015). Similarly, the difference in engine power, and thus in fuel consumption, between Ice Class and conventional container vessels was statistically significant. The results indicate that, generally, engine power, fuel consumption and fuel costs are around 10% higher for Ice Class IA/IAS vessels than for conventional vessels. This figure exceeds previous estimates given by Dvorak (2009). Our statistical analysis revealed no significant differences between Ice Class and conventional vessels w.r.t.

T. Solakivi et al.

the effects of the Ice Class on the main-engine SFOC. Again, this result challenges previous findings reported by Furuichi and Otsuka (2015). It has been assumed that the operating costs of Ice Class IA/IAS and conventional vessels are identical. Our results indicate that the total shipping costs (including capital, operating and fuel costs) of IA/IAS vessels are, on average, 9% higher than those of conventional vessels. This is slightly more than 1 USD per day per TEU. As such, this could be considered as the additional cost of operating with an Ice Class fleet during the open-water period. However, it could also be considered as the lower boundary of the real cost differential, which only concerns the technical details, and excludes the effects of winter conditions in the form of higher real fuel consumption and a lower operating speed. Ship owners operating in the Baltic Sea estimate that the increase in fuel consumption in ice conditions during the winter may be as high as 30%. At the same time, the operating speed of the vessel may be as much as 30% lower than in open waters. This also results in higher costs for shippers in terms of capital tied up in inventories, as Maloni et al. (2013) suggest. Given these estimates, the ‘‘Cost of Ice’’ during the winter could be as high as 30%, or over USD 4/TEU/day for a container vessel of 4000 TEU. The cost differential would be similar for a 10,000 TEU vessel, indicating that some previous results on the feasibility of Arctic Sea routes (Xu et al. 2011) may be over-optimistic. Our results should be interpreted with caution. Although technical data may be objective, differences in the data, including the variance in engine power and new building prices, may be subject to variables beyond this analysis. For instance, vessels equipped with more powerful engines may be able to operate with a lower engine load in certain conditions, thus affecting fuel consumption. Engine power may be partly attributable to the design speed of a vessel, planned for a specific route, for example. At the same time, the high R2s of the estimated models indicate that the included variables capture the essentials. Capital and fuel costs are also highly sensitive to interest rates and fuel prices, respectively. The current state of the global economy has kept both on a low level, even when average values, over several years, are used. This paper shows clearly that the requirements of the operating environment should be considered in studies on shipping economics, especially in extreme weather conditions. IA and IAS Ice Class vessels designed for navigation in first-year ice represent a compromise between winter and summer requirements, thereby providing a sub-optimal solution for both. Ship owners intending to navigate in winter ice conditions are burdened with additional costs during the rest of the year. The novelty of our research is in distinguishing Ice Class vessels as a separate entity. There are numerous countries in the Northern Hemisphere for which foreign trade is highly dependent on uninterrupted maritime transport, even in winter, and an Ice Class fleet serves as a key conveyor of foreign trade for them (Koskinen and Hilmola 2005). On the macro level, the results of our research carry solid policy implications, in terms of balancing the gap in logistics competitiveness. The shipping industry will also benefit from the provision of quantitative information, to facilitate their fleet-investment and management decisions.

On the cost of ice: estimating the premium of Ice Class…

From a wider perspective, our results could be used as a baseline on which to assess the feasibility of Arctic Sea routes. Even if the existing literature on this is by no means scarce, it nevertheless offers diverse conclusions (Lasserre 2014). The technical requirements (including ice-breaking services and fees) for the utilisation of the Northern Sea Route on a seasonal basis are consistent with the requirements of the IA and IAS Ice Classes (Gritsenko and Kiiski 2016; Kiiski et al. 2016). Thus the cost estimates of this research may be used as a lower limit for the true costs of utilising Arctic Sea routes. The results of our research apply only to containerships. Similar methodology could be applied to other types of vessels to produce an overall picture of the ‘‘Cost of Ice’’ in shipping. The results also provide a starting point for more detailed analysis of the environmental effects of shipping. Previous research on emissions is strongly based on the assumption of an average vessel in open-water conditions (Corbett and Koehler 2003; Corbett et al. 2009). Given the additional engine power of Ice Class vessels and the seasonal increase in fuel consumption, shipping emissions in winter conditions are considerably underestimated. There is thus a need for more detailed analysis of the environmental impact of shipping and its emissions. The unresolved contradiction between EEDI requirements for energy efficiency and the need for extra propulsion power in ice-infested waters warrants particular attention. Acknowledgements The authors would like to thank the anonymous reviewers and the editor-in-chief for their valuable comments in developing this paper. We would also like to acknowledge the Jenny and Antti Wihuri Foundation for their financial support.

References Bunker Index. 2015. Available at http://www.bunkerindex.com/. Accessed June 2015. Bourbonnais, P., and F. Lasserre. 2015. Winter Shipping in the Canadian Arctic: Toward Year-Round Traffic? Polar Geography 38 (1): 70–88. CRSL; Clarkson Research Services Limited. 2015. Clarksons World Fleet Register. https://www. clarksons.net/wfr2/. Accessed Jan 2015. Cullinane, K., and M. Khanna. 1999. Economies of Scale in Large Container Ships. Journal of Transport Economics Policy 33 (2): 185–208. Corbett, J.J., and H.W. Koehler. 2003. Updated Emissions from Ocean Shipping. Journal of Geophysical Research 108 (D20): 4650. Corbett, J.J., H. Wang, and J.J. Winebrake. 2009. The Effectiveness and Costs of Speed Reductions on Emissions from International Shipping. Transportation Research Part D 14 (8): 593–598. Draper, N.R., and H. Smith. 1998. Applied Regression Analysis. New York: Wiley. Drewry. 2012. Ship Operating Costs Annual Review and Forecast: Annual Report 2012/13. London: Drewry Shipping Consultants. Dvorak, R.E. 2009. Engineering and Economic Implications of Ice-Classed Containerships. MSc Dissertation. Massachusetts Institute of Technology, Cambridge, MA. Erikstad, S.O., and S. Ehlers. 2012. Decision Support Framework for Exploiting Northern Sea Route Transport Opportunities. Ship Technology Research 59 (2): 34–42. Federal Reserve. 2016. Foreign exchange rates, Board of Governors of the Federal Reserve System. https://www.federalreserve.gov/releases/h10/hist/. Retrieved 21 Apr 2015. Finnish Transport Safety Agency. 2011. Guidelines for the application of the Finnish-Swedish ice class rules. TRAFI/21816/03.04.01.01/2011 Helsinki, 20 December, 2011. Furuichi, M., and N. Otsuka. 2015. Proposing a Common Platform of Shipping Cost Analysis of the Northern Sea Route and the Suez Canal Route. Maritime Economics and Logistics 17 (1): 9–31.

T. Solakivi et al. Goerlandt, F., J. Montewka, W. Zhang, and P. Kujala. 2016. An Analysis of Ship Escort and Convoy Operations in Ice Conditions. Safety Science 95: 198–209. Gritsenko, D., and T. Kiiski. 2016. A Review of Russian Ice-Breaking Tariff Policy on the Northern Sea Route 1991–2014. Polar Record 52 (2): 144–158. Hummels, D. 2007. Transportation Costs and International Trade in the Second Era of Globalization. The Journal of Economic Perspectives 21 (3): 131–154. Kiiski, T., T. Solakivi, J. To¨yli, and L. Ojala. 2016. Long-Term Dynamics of Shipping and Icebreaker Capacity Along the Northern Sea Route. Maritime Economics & Logistics. doi:10.1057/s41278-0160049-1. Koskinen, M.-M., and O.-P. Hilmola. 2005. Investment Cycles in the Newbuilding Market of IceStrengthened Oil Tankers. Maritime Economics & Logistics 7 (2): 173–188. Krugman, P. 1991. History and Industry Location: the Case of Manufacturing Belt. The American Economic Review 81 (2): 80–83. Laulajainen, R. 2009. The Arctic Sea Route. International Journal of Shipping and Transport Logistics 1 (1): 55–73. Lasserre, F. 2014. Case Studies of Shipping Along Arctic Routes. Analysis and profitability perspectives for the container sector. Transportation Research Part A 66: 144–161. Liu, M., and J. Kronbak. 2010. The Potential Economic Viability of Using the Northern Sea Route (NSR) as an Alternative Route Between Asia and Europe. Journal of Transport Geography 18 (3): 434–444. Lloyd’s Register. 2015. LR to class 3600 TEU Container Ships for Maersk Line—First Boxships to be Built at COSCO Zhoushan. http://www.lr.org/en/news/news/LR-to-class-3600-teu-container-shipsfor-Maersk-Line-first-boxships-to-be-built-at-COSCO-Zhoushan.aspx. Accessed 28 Oct 2015. Maersk. 2015. Maersk Line Orders Seven Ice-Class Container Vessels. http://www.maerskline.com/ro-ro/ countries/int/news/news-articles/2015/03/seven-ice-class-container-vessels. Accessed 28 Oct 2015. Maloni, M., J.A. Paul, and D.M. Gligor. 2013. Slow Steaming Impacts on Ocean Carriers and Shippers. Maritime Economics and Logistics 15 (2): 151–171. MAN Diesel & Turbo. 2013. Ice Classed Ships. Copenhagen: MAN Diesel & Turbo. Omstedt, A., J. Elken, A. Lehmann, M. Leppa¨ranta, H.M.E. Meier, K. Myrberg, and A. Rutgersson. 2014. Progress in Physical Oceanography of the Baltic Sea During the 2003–2014 Period. Progress in Oceanography 128: 139–171. Pruyn, J.F.J. 2016. Will the Northern Sea Route Ever be a Viable Alternative? Maritime Policy & Management 43 (6): 661–675. Riska, K., and J. Ka¨ma¨ra¨inen. 2011. A Review of Ice Loading and the Evolution of the Finnish-Swedish Ice Class Rules. Transactions of the Society of Naval Architects and Marine Engineers 119: 265–298. Riska, K., Wilhelmson, M., Englund, K. (1997) Performance of merchant vessels in ice in the Baltic, Finnish Maritime Administration, Winter Navigation Research Board, Research Report No. 52. Stephenson, S.R., L.C. Smith, L.W. Brigham, and J.A. Agnew. 2013. Projected 21st-Century Changes to Arctic Marine Access. Climate Change 118 (3–4): 885–899. Tran, N.K., and H.-D. Haasis. 2015. An Empirical Study of Fleet Expansion and Growth of Ship Size in Container Liner Shipping. International Journal of Production Economics 159: 241–253. UNCTAD—United Nations Conference on Trade and Development. 2016. Review of Maritime Transport 2016. New York: United Nations Publications. Wijnolst, N., and T. Wergeland. 2009. Shipping Innovation. Amsterdam: IOS Press. World Shipping Council. 2015. http://www.worldshipping.org/. Accessed 27 Oct 2015. WTO. 2015. World Trade Organization Statistics Database. Xu, M., Z. Yin, D. Jia, F. Jin, and H. Ouyang. 2011. The Potential Seasonal Alternative of Asia-Europe Container Service Via Northern Sea Route Under the Arctic Sea Ice Retreat. Maritime Policy & Management 38 (5): 541–560.